Research method for fire endurance after impact of steel tube confined concrete

Abstract

The invention relates to the technical field of fire resistance detection after impact of a concrete wall, in particular to a fire endurance research method after impact of steel tube confined concrete. The method comprises the steps that a finite element impact simulation model and a static analysis model are established, and the maximum impact energy critical value capable of being borne by a component is obtained; calculating a component temperature field distribution condition through heat transfer analysis; obtaining the initial fire endurance of the component from the initial fire endurance model; obtaining a damaged state after impact; calculating the fire endurance of the component under the combined action of impact and fire; and establishing a machine learning model. According to the method, a general machine learning prediction model is established through multiple types of component parameters, the dependence of a traditional test method on a specific component form is avoided, the evaluation cost is remarkably reduced, meanwhile, through correlation modeling of impact damage and fire resistance, under the condition of lack of test data, the fire resistance limit damage condition after impact can still be simulated, and the evaluation accuracy is improved. And a new method is provided for compound disaster research.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of post-impact fire resistance detection of concrete walls, and in particular, to a method for researching the post-impact fire resistance limit of steel pipe confined concrete. BACKGROUND

[0002] There are mainly two ways to study the damage condition of components. One is a real test in the laboratory, which reproduces the damage condition that the component may suffer by simulating the disaster scene. The other is through the method of finite element simulation. The first step is to establish the constitutive model of the material, and then to build a finite element model. With the help of professional finite element analysis software, we can accurately calculate the damage result of the component under certain conditions.

[0003] In the early research of scholars in various countries, the impact and fire on the structure were discussed respectively. However, due to the completely different damage mechanisms of impact load and high temperature of fire on the structure, the research on the combined action of the two is relatively less. Many events make people realize that the combined action of impact and fire will cause serious damage to the structure. Therefore, studying the damage mechanism and safety evaluation of buildings subjected to the combined action of impact load and fire, and putting forward measures to improve the fire resistance and impact resistance of structures, has become an important problem that the engineering field urgently needs to solve, and has important scientific and engineering significance.

[0004] The present application is based on a calculation method combining finite element and machine learning. Its significant advantage lies in the combination of the accurate calculation ability of finite element analysis and the prediction accuracy and efficiency of machine learning model. This method can quickly and accurately predict the damage condition under the premise of ensuring calculation accuracy, thereby effectively avoiding the risks and resource consumption that may be brought by real test. At the same time, it also greatly reduces the dependence of finite element simulation on computing resources, reduces the professional threshold of operation, and makes the whole prediction process more efficient and easy to implement.

[0005] A method combining finite element simulation and machine learning is used to study the steel pipe confined concrete column. This method analyzes the attenuation degree of the fire resistance limit of the component after being subjected to the combined action of impact and fire, to accurately evaluate its load bearing capacity. The advantage is that it can accurately and quickly predict the load bearing requirement of the component after being subjected to the combined action of impact and fire. SUMMARY

[0006] The purpose of the present application is to provide a method for researching the post-impact fire resistance limit of steel pipe confined concrete, which can quickly and accurately predict the fire resistance limit of the component after being subjected to the combined action of impact and fire, and make its effectiveness meet the future judgment and design optimization method of steel pipe confined concrete column damage.

[0007] To achieve the above object, a method for researching post-impact fire resistance limit of steel tube confined concrete is provided, comprising the following steps:

[0008] S1, a finite element impact simulation model is established, a static analysis model is established by using a finite element software ABAQUS, an axial load value is applied to the component in the finite element impact model according to the value of the axial compression ratio, the post-impact deflection is selected as the standard for judging whether the component reaches the failure state, and the maximum impact energy critical value that the component can withstand is obtained accordingly;

[0009] S2, a temperature field model is established by using a heat transfer analysis step, the temperature field distribution of the component is calculated based on the ISO-834 standard temperature rising curve through heat transfer analysis;

[0010] S3, an initial fire resistance limit model is established by using a static general analysis step, the initial fire resistance limit of the component is obtained from the initial fire resistance limit model;

[0011] S4, the post-impact fire resistance limit is obtained: the impact working conditions are designed according to 20%, 40% and 60% of the maximum impact energy critical value obtained in S1, and the damaged state after impact is obtained;

[0012] S5, the impact damage state is imported into the initial fire resistance limit model in S3 as the initial state, and the fire resistance limit of the component under the combined action of impact and fire is calculated;

[0013] S6, a machine learning model is established according to the method of building a stacked ensemble.

[0014] As a further improvement of the technical solution, the specific steps of S1 are:

[0015] a, a static analysis model is established by using a finite element software ABAQUS, the initial bearing capacity of the component is obtained by applying displacement, the axial load value is obtained according to the axial compression ratio, and the axial load is applied in the impact model and the fire resistance limit model;

[0016] b, an impact model is established by using explicit dynamics, the mass of the impact body is given, the distance between the impact body and the component is set, the speed of the impact body is set in the pre-defined field, and the impact event simulation is performed;

[0017] c, the steel tube-concrete interface Coulomb friction model and the impact body-steel tube interface are set;

[0018] d, the continuous geometric model is divided into many small grids;

[0019] e, the failure displacement of the component is determined according to the maximum displacement criterion, and the maximum impact energy critical value that the component can withstand is obtained.

[0020] As a further improvement of the technical solution, the S2 calculates the temperature rise of the component under the ISO-834 standard temperature rise curve before the ISO-834 standard temperature rise curve, and obtains the temperature rise process of the component and the node temperature by using the heat transfer analysis step, the surface heat exchange condition, the surface radiation and the heat conduction.

[0021] As a further improvement of the technical solution, in the S3, when establishing the initial fire resistance limit model, a model consistent with the temperature field model component size and grid division is established, a static general analysis step is adopted, a load is applied according to the axial compression ratio value, and the temperature field calculation result data is imported into the temperature predefined field, and the "displacement-time" curve extracted from the load point is taken as the fire resistance limit result.

[0022] As a further improvement of the technical solution, the step of calculating the fire resistance limit of the component in the S5 is:

[0023] The damaged state after impact is imported into the fire resistance limit model as the initial state;

[0024] The impact damaged state is imported into the Initial analysis step predefined field as the initial state, and the impact damaged state is imported into the Initial predefined field of the initial fire resistance limit model as the initial state;

[0025] The calculated temperature field model result data is imported into the Step-1 analysis step temperature predefined field;

[0026] The temperature field result is imported into the Step-1 analysis step predefined field, and the load is applied in the Step-1 analysis step, and then the post-impact fire resistance limit value is obtained.

[0027] As a further improvement of the technical solution, the grid division of the post-impact fire resistance limit model in the S5 is consistent with the impact model and the temperature field model, and the boundary condition is consistent with the impact model.

[0028] As a further improvement of the technical solution, the method for establishing the machine learning model in the S6 is:

[0029] A stacking regression method is used, three CatBoost models are used as base learners, a PSO particle swarm optimization algorithm is used for hyperparameter optimization, a LightGBM model is used as a meta-learner, and SHAP analysis parameter importance and key parameter coupling are used to establish the machine learning model.

[0030] As a further improvement of the technical solution, in the S6, when SHAP is analyzed, the SHAP value of each feature in all samples is calculated, the mean of the absolute value of each feature is taken, and the mean is sorted from high to low to obtain the feature importance ranking.

[0031] Compared with the prior art, the present application has the beneficial effects of:

[0032] 1. In the method for researching the post-impact fire resistance limit of steel pipe confined concrete, a general machine learning prediction model is established through multiple types of component parameters, avoiding the dependence of traditional test methods on specific component forms, significantly reducing the evaluation cost, and at the same time, through the modeling of the correlation between impact damage and fire resistance performance, the post-impact fire resistance limit damage can still be simulated in the absence of test data, providing a new method for composite disaster research. Through SHAP analysis, the sensitivity of parameters such as impact energy percentage and axial compression ratio on fire resistance limit is determined, guiding the design stage of impact-resistant-fire-resistant collaborative optimization and improving the safety of the structure throughout its life cycle. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 The figure is a schematic diagram of the overall workflow of the present application;

[0034] Figure 2 The figure is a schematic diagram of the impact simulation finite element calculation of the steel pipe confined concrete column of the present application;

[0035] Figure 3 The figure is a schematic diagram of the temperature field finite element calculation of the steel pipe confined concrete column of the present application;

[0036] Figure 4 The figure is a schematic diagram of the fire resistance limit finite element calculation of the steel pipe confined concrete column of the present application;

[0037] Figure 5 The figure is a schematic diagram of the post-impact fire resistance limit finite element calculation method of the steel pipe confined concrete column of the present application;

[0038] Figure 6 The figure is a flowchart of the post-impact fire resistance limit data set loading process of the present application;

[0039] Figure 7 The figure is a CatBoost-LGBM model architecture diagram of the present application. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0041] In the description of the present application, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application.

[0042] Referring to Figures 1-7 The present application provides a method for researching the post-impact fire resistance limit of steel pipe confined concrete, comprising the following steps:

[0043] S1, establishing a finite element impact simulation model, using finite element software ABAQUS to establish a static analysis model, obtaining the initial bearing capacity of the steel pipe confined concrete column, and according to the value of the axial compression ratio, applying an axial load value to the component in the finite element impact model, establishing and verifying the core concrete constitutive model (stress-strain relationship), obtaining the impact damaged state of the concrete, and comparing and analyzing with the existing experimental data, selecting the post-impact deflection as the standard for judging whether the component reaches the failure state, and obtaining the maximum impact energy critical value that the component can withstand according to the above;

[0044] The specific steps are:

[0045] a, using finite element software ABAQUS to establish a static analysis model, obtaining the initial bearing capacity by applying displacement, obtaining the axial load value according to the axial compression ratio, and applying the axial load in the impact model and the fire resistance limit model;

[0046] The axial compression ratio is the ratio of the axial load value to the initial bearing capacity. First, the axial compression ratio is determined;

[0047] For example, 0.4, and then the axial load value is obtained according to the known initial bearing capacity x 0.4;

[0048] b, using explicit dynamics to establish an impact model, by assigning the mass of the impact body, setting the distance between the impact body and the component, setting the impact body velocity in the pre-defined field, and simulating the impact event;

[0049] The steel pipe and the concrete are assigned C3D8R mesh, the built-in steel cage is assigned T3D4 mesh, and the impact body is assigned R3D4 mesh;

[0050] c, setting the steel pipe-concrete interface Coulomb friction model and the impact body-steel pipe interface;

[0051] The interaction between the inner surface of the steel pipe and the outer surface of the concrete is modeled by Coulomb friction, with a tangential friction coefficient of 0.6 and normal hard contact; the steel cage is set as an internal region, with the main body being the core concrete; the interaction between the lower surface of the impact body and the outer surface of the steel pipe is also modeled by Coulomb friction, with a tangential friction coefficient of 0.2 and normal hard contact; at one end of the component where the load is applied, rotation in all directions except the z direction is restricted, and at the other end, rotation in all directions is restricted;

[0052] d. In order to facilitate computer calculation, the continuous geometric model is divided into many small grids; different grid properties are suitable for different calculations, and in this step, the mechanical properties are mainly utilized, so the C3D8R (steel pipe and concrete), T3D4 (steel cage), and R3D4 (impact body) grid properties are assigned;

[0053] e. The failure displacement of the component is determined according to the maximum displacement criterion, and the maximum impact energy critical value that the component can withstand is obtained; when the deflection reaches 1 / 12 of the length, it is considered to be failed, and the deflection value is taken as the criterion;

[0054] The maximum displacement criterion is the standard for judging the failure of the component, and the displacement is extracted from the post-impact result file.

[0055] S2, a temperature field model is established using the heat transfer analysis step, based on the ISO-834 standard temperature rise curve, the temperature field distribution of the component is calculated through heat transfer analysis;

[0056] The temperature rise of the component under the ISO-834 standard temperature rise curve is calculated, and the temperature rise process and node temperature of the component are obtained using the heat transfer analysis step, surface heat exchange conditions, surface radiation, and thermal conduction;

[0057] In the temperature field analysis, the Stefan-Boltzmann constant is set to 5.67e-08, and the absolute zero is set to -273.13; in this step, the thermal properties are mainly involved, so the steel pipe, concrete, and steel cage grid properties are assigned DC3D8 grid properties, the grid division is consistent with the impact model, and binding constraints and thermal conduction are set between the steel pipe and the concrete; the analysis step time is the specified fire time;

[0058] S3, an initial fire resistance limit model is established using the static general analysis step, and the initial fire resistance limit of the component is obtained from the initial fire resistance limit model;

[0059] A model consistent with the component size and grid division of the temperature field model is established, the static general analysis step is used, the load is applied according to the axial compression ratio value, and the temperature field calculation result data is imported into the temperature predefined field; the "displacement-time" curve extracted from the load point is taken as the fire resistance limit result;

[0060] In this step, the steel pipe and the concrete grid attribute are set to C3D8R, and the reinforcement cage is set to T3D4; the analysis step time and the temperature field analysis time are consistent, the load is applied in the Step-1 analysis step, and the temperature field calculation result is imported in the predefined field, so as to obtain the initial fire resistance limit of the component;

[0061] The displacement-time curve of the load point can identify the time point at which the component loses the carrying capacity or reaches the critical failure for the first time in the fire condition. Therefore, the maximum value of the displacement-time curve of the load point is the initial fire resistance limit;

[0062] S4, obtain the post-impact fire resistance limit: design impact working conditions according to 20%, 40%, and 60% of the maximum impact energy critical value obtained in S1 to obtain the damaged state after impact;

[0063] S5, import the impact damage state as the initial state into the initial fire resistance limit model in S3 to calculate the fire resistance limit of the component under the combined action of impact and fire;

[0064] The steps of post-impact fire resistance limit calculation are: importing the damaged state after impact as the initial state into the fire resistance limit model; importing the calculated temperature field model result data into the Step-1 analysis step temperature predefined field to achieve the purpose of post-impact fire resistance limit calculation;

[0065] The grid division of the post-impact fire resistance limit model is consistent with the impact model and the temperature field model, the boundary conditions are consistent with the impact model, the impact damaged state is imported as the initial state in the Initial analysis step predefined field, the impact damaged state is imported as the initial state in the Initial predefined field of the initial fire resistance limit model, the temperature field result is imported in the Step-1 analysis step predefined field, and the load is applied in the Step-1 analysis step, and then the post-impact fire resistance limit value is obtained, and the fire resistance limit value is the maximum value of the time-displacement curve;

[0066] The odb result calculation file automatically generated after ABAQUS calculation contains the impact damaged state; in ABAQUS, the analysis should follow the order of the analysis steps, and the Initial analysis step is calculated first; the Initial analysis step predefined field is a setting under this analysis step.

[0067] S6, establish a machine learning model according to the building stacking integration method;

[0068] Method for establishing a machine learning model:

[0069] The three CatBoost models are used as base learners, and the PSO particle swarm optimization algorithm is used for hyperparameter tuning. The LightGBM model is used as a meta-learner. The three CatBoost models are used as base learners, the LightGBM is used as a meta-learner, and the PSO algorithm is used to optimize the hyperparameters. The SHAP analysis parameter importance and key parameter coupling are combined.

[0070] The CatBoost base learner uses a symmetric tree structure and ordered boosting technology to solve prediction bias, enhance robustness, and avoid overfitting. The three CatBoost base learners are adjusted by the PSO particle swarm optimization algorithm to provide more accurate feature representation and prediction output, laying a solid foundation for the meta-learner. The PSO particle swarm optimization algorithm is used to adjust the hyperparameters of the CatBoost model. The LightGBM meta-learner uses histogram decision trees and one-sided gradient sampling technology to improve training speed and save memory. The LightGBM is selected as the meta-learner, which can efficiently integrate the prediction results of multiple CatBoost base learners, and further improve the generalization ability and prediction accuracy of the model using its fast and low-consumption advantages. Based on the model, SHAP parameter analysis is performed to obtain the contribution of each parameter to the fire resistance limit and the coupling mechanism between parameters;

[0071] Based on the model, SHAP model explainable analysis is performed to explore the key parameters affecting the fire resistance limit of the component after impact and the coupling between parameters;

[0072] The SHAP analysis tool can quantify the contribution of each parameter to the model output through global feature importance analysis and identify the main influencing factors. The interaction effect analysis reveals the coupling mechanism between parameters. Based on the parameter analysis results, guidance suggestions are provided for engineering design.

[0073] The SHAP analysis tool calculates the SHAP value of each feature in all samples, takes the mean of the absolute value of each feature, and sorts the mean values from high to low to obtain the feature importance ranking.

[0074] The input parameters of the established machine learning model dataset include: concrete strength, steel yield strength, steel pipe wall thickness, radius, length, axial compression ratio, and impact energy percentage value. The output parameter is the fire resistance limit prediction value.

[0075] Fire resistance limit prediction value calculation method: the model analyzes the rules in historical data, establishes a mapping relationship between input parameters and fire resistance limit, and calculates the result according to this rule when encountering new data. The essence is to extract patterns from input data through mathematical models and map them to target answers.

[0076] Example 1

[0077] The present application takes the steel pipe confined concrete column with the size in Table 1 as the research object, and the application is described in detail in combination with the drawings.

[0078] Table 1

[0079]

[0080] S1, first, the static constitutive material of concrete and steel under the size and strength shown in Table 1 is determined, the static general analysis step in ABAQUS software is used to apply displacement at the top end of the column, and the force-displacement curve is obtained after the calculation is completed, and the maximum value is the initial bearing capacity of the component. The initial bearing capacity of the component shown in Table 1 is 6424420 Mpa; according to the value of the axial compression ratio of 0.7, the axial load applied in the impact calculation model is calculated as 6424420*0.7=4497094 Mpa; then the dynamic constitutive material of concrete and steel considering the strain rate effect under the size and strength shown in Table 1 is determined, the impact finite element analysis model is established by means of the display dynamic analysis step, and the axial load of 4497094 Mpa is applied to the component; according to the maximum displacement criterion, the deflection after impact reaching 1 / 12 of the length of the component is regarded as the destruction of the component, so the deflection of the component when reaching the destruction is 300 mm, and the maximum impact energy critical value that the component can withstand is 1000000 J at this time;

[0081] S2, the temperature field model is established by means of the heat transfer analysis step, the ISO-834 standard temperature rising curve is used to simulate the temperature rising process; the Stefan-Boltzmann constant is 5.67e-08, and the absolute zero is-273.13; the “Tie” constraint is established between the steel pipe and the concrete; the thermal radiation and the thermal convection are set on the surface of the steel pipe; the grid division is consistent with the impact model, and the node temperature and the temperature rising condition of the component are obtained;

[0082] S3, the initial fire resistance limit model is established by means of the static general analysis step, the model components are consistent with the temperature field model; the grid division is consistent with the impact model of S1, the temperature field model of S2, and the temperature field result calculated in S2 is imported into the Step-1 analysis step pre-defined field-temperature, and the axial load of 4497094 Mpa is applied, so that the initial fire resistance limit value is obtained;

[0083] S4, according to S1, the maximum impact energy critical value of the component is 1000000 J, 20%, 40% and 60% of the maximum impact energy critical value are taken, i.e. 200000 J, 400000 J and 600000 J, the impact working condition is designed according to the impact energy calculation formula 1 / 2mv2, and the impact working condition value in the impact model of S1 is modified to obtain the damaged state after impact;

[0084] S5, using the initial fire resistance limit model in S3, importing the impact damage calculation result odb file into the initial analysis step-initial state, that is, the post-impact fire resistance limit calculation is realized;

[0085] S6, record the initial fire resistance limit value obtained in S3 and the post-impact fire resistance limit value obtained in S5 in the data set, as shown in the following table:

[0086] Table 2

[0087]

[0088] The CatBoost-LGBM model is used for prediction, and the fire resistance limit prediction value in Table 2 is referenced in Table 3:

[0089] Table 3

[0090]

[0091] Example 2

[0092] Now taking the steel pipe confined concrete column with the size of Table 4 as the research object, the application will be described in detail in combination with the drawings.

[0093] Table 4

[0094]

[0095] S1, first determine the static constitutive material of concrete and steel under the size and strength shown in Table 4, apply displacement at the top end of the column by means of the static general analysis step in ABAQUS software, obtain the force-displacement curve after the calculation is completed, and the maximum value is the initial bearing capacity of the component, the initial bearing capacity of the component shown in Table 4 is 2860840Mpa; according to the value of the axial compression ratio 0.6, the axial load applied in the impact calculation model is calculated as 2860840*0.6=1716504Mpa; then determine the dynamic constitutive material of concrete and steel considering strain rate effect under the size and strength shown in Table 4, establish an impact finite element analysis model by means of the display dynamic analysis step, and apply an axial load of 1716504Mpa to the component; according to the maximum displacement criterion, the deflection after impact reaching 1 / 12 of the length of the component is regarded as the destruction of the component, so the deflection of the component when reaching the destruction is 133mm, and the maximum impact energy threshold that the component can withstand is 160000J at this time;

[0096] S2, a temperature field model is established by using heat transfer analysis step, an ISO-834 standard temperature rising curve is used to simulate the temperature rising process; a Stefan-Boltzmann constant is 5.67e-08, and an absolute zero is -273.13; a “Tie” constraint is established between the steel pipe and the concrete; heat radiation and heat convection are set on the surface of the steel pipe; mesh division is consistent with the impact model, and node temperature and temperature rising of the component are obtained;

[0097] S3, an initial fire resistance limit model is established by using static general analysis step, model components are consistent with the temperature field model; mesh division is consistent with the impact model of S1 and the temperature field model of S2; the temperature field result calculated in S2 is imported into Step-1 analysis step pre-defined field-temperature, and an axial load of 1716504Mpa is applied;

[0098] S4, according to S1, the maximum impact energy critical value of the component is 160000J, 20%, 40% and 60% of the maximum impact energy critical value are taken, that is, 32000J, 64000J and 96000J, impact working conditions are designed according to the formula of impact energy = 1 / 2mv2, and the impact working condition value in the impact model of S1 is modified to obtain the damaged state after impact;

[0099] S5, the initial fire resistance limit model in S3 is used, the impact damage calculation result odb file is imported into Initial analysis step-initial state, and the post-impact fire resistance limit calculation is realized;

[0100] S6, the initial fire resistance limit value obtained in S3 and the post-impact fire resistance limit value obtained in S5 are recorded in a data set, as shown in Table 5:

[0101] Table 5

[0102]

[0103] The CatBoost-LGBM model is used for prediction, and the fire resistance limit prediction value in Table 5 is shown in Table 6:

[0104] Table 6

[0105]

[0106] Example 3

[0107] The following will take an example to calculate another case, and the component with the size of Table 7 is taken as the research object, and the application will be described in detail with reference to the accompanying drawings.

[0108] Table 7

[0109]

[0110] S1, first determine the static constitutive material of concrete and steel under the size and strength shown in Table 7, by means of the static general analysis step in ABAQUS software, the displacement is applied at the top of the column, and the force-displacement curve is obtained after the calculation is completed, the maximum value is the initial bearing capacity of the component, the initial bearing capacity of the component shown in Table 7 is 49094900Mpa; According to the value of axial compression ratio 0.6, the axial load applied in the impact calculation model is calculated as 49094900*0.4=19637960Mpa; Then determine the dynamic constitutive material of concrete and steel considering strain rate effect under the size and strength shown in Table 7, establish the impact finite element analysis model by means of the display dynamic analysis step, and apply an axial load of 19637960Mpa to the component; According to the maximum displacement criterion, the deflection after impact reaches 1 / 12 of the length of the component, which is regarded as the destruction of the component, so the deflection of the component when it reaches the destruction is 375mm, and the maximum impact energy threshold that the component can withstand is 9600000J at this time;

[0111] S2, establish a temperature field model by using the heat transfer analysis step, and simulate the temperature rising process by ISO-834 standard temperature rising curve; Stefan-Boltzmann constant is 5.67e-08, and absolute zero is-273.13; Establish "Tie" constraint between steel pipe and concrete; Set up thermal radiation and convection on the surface of steel pipe; The grid division is consistent with the impact model, and the node temperature and temperature rising of the component are obtained;

[0112] S3, establish the initial fire resistance limit model by using the static general analysis step, the model components are consistent with the temperature field model; The grid division is consistent with the impact model of S1 and the temperature field model of S2; The temperature field results calculated in S2 are imported into Step-1 analysis step pre-defined field-temperature, and an axial load of 19637960Mpa is applied;

[0113] S4, according to S1, the maximum impact energy threshold of the component is 9600000J, take 20%, 40%, 60% of the maximum impact energy threshold respectively, that is, 1920000J, 3840000J, 5760000J, design the impact working condition according to the formula of impact energy=1 / 2mv2, and modify the impact working condition value in S1 impact model, get the damaged state after impact;

[0114] S5, use the initial fire resistance limit model in S3, import the impact damaged calculation result odb file into the initial state of Initial analysis step, which can realize the calculation of post-impact fire resistance limit;

[0115] S6, record the initial fire resistance limit value obtained by S3 and the post-impact fire resistance limit value obtained by S5 in the data set, as shown in Table 8:

[0116] Table 8

[0117]

[0118] The fire endurance prediction values in Table 8 using the CatBoost-LGBM model are shown in Table 9:

[0119] Table 9

[0120]

[0121] According to the SHAP analysis results, the axial compression ratio has the highest SHAP value and is the most important input feature in fire endurance prediction, followed by core concrete importance, concrete strength, steel yield strength, and wall thickness. When the impact energy is zero, the wall thickness has a significant weakening effect on the fire endurance, but as the impact energy increases, this weakening effect gradually weakens. This is because the increase in wall thickness can improve the impact resistance of the component, so that its negative impact is partially offset under subsequent high temperature load. Therefore, in the analysis of post-impact fire endurance, the SHAP value of the wall thickness is relatively low.

[0122] As the impact energy increases, the contribution of each parameter to the fire endurance decreases, with the axial compression ratio and radius being the most obvious. The axial compression ratio has a negative effect on the impact resistance and fire endurance of the component, and its influence on post-impact fire endurance gradually dominates, but as the impact energy increases, its dominant role is weakened. The increase in component radius significantly improves its impact resistance, and the increase in core concrete proportion enhances the fire resistance of the component. However, as the impact energy continues to rise, the performance improvement brought by these factors tends to weaken, and the enhancement effect gradually decreases.

[0123] When the impact energy reaches 40%, the weakening effect of the axial compression ratio on the fire endurance of the component and the enhancement effect of the cross-sectional radius on it both show an accelerated downward trend, indicating that the overall performance of the component has degraded and its disaster resistance has weakened significantly. This phenomenon shows that at a moderate impact energy level, the regulation of each parameter on the fire performance of the component begins to weaken, and external impact gradually dominates. When the impact energy reaches 60%, the damage caused by the impact has already significantly exceeded the regulation ability of the component's own parameters on the fire endurance, resulting in a decrease in the sensitivity of the component to parameter changes. At this time, improving the impact resistance of the component becomes the key to ensuring its fire endurance.

[0124] Through the multi-type component parameters, a general machine learning prediction model is established, avoiding the dependence of traditional test methods on specific component forms, significantly reducing the evaluation cost, and through the modeling of the correlation between impact damage and fire resistance performance, the damage condition of the fire resistance limit after impact can still be simulated in the absence of test data, providing a new method for composite disaster research. Through SHAP analysis, the sensitivity of parameters such as impact energy percentage and axial pressure ratio to the fire resistance limit is determined, guiding the design stage of impact-fire resistance collaborative optimization and improving the safety of the structure in the whole life cycle.

[0125] The above shows and describes the basic principles, main features and advantages of the present application. Those skilled in the art should understand that the present application is not limited to the above-mentioned embodiments, and the above-mentioned embodiments and descriptions in the specification are only preferred examples of the present application and are not intended to limit the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for studying the fire resistance limit of steel-tube confined concrete after impact, characterized in that: Includes the following steps: S1. Establish a finite element impact simulation model. Use the finite element software ABAQUS to establish a static analysis model. Apply axial load values ​​to the component in the finite element impact model according to the axial compression ratio. Select the deflection after impact as the standard for judging whether the component has reached the failure state. Based on this, obtain the critical value of the maximum impact energy that the component can withstand. S2. Establish a temperature field model using heat transfer analysis. Based on the ISO-834 standard heating curve, calculate the temperature field distribution of the component through heat transfer analysis. S3. Use the static general analysis step to establish the initial fire resistance limit model, and obtain the initial fire resistance limit of the component from the initial fire resistance limit model; S4. Obtain the fire resistance limit after impact: Design impact conditions at 20%, 40%, and 60% of the maximum impact energy critical value obtained in S1, and obtain the damage state after impact. S5. Import the impact damage state as the initial state into the initial fire resistance limit model in S3, and calculate the fire resistance limit of the component under the combined action of impact and fire. S6. Based on the construction stacking ensemble method, establish a machine learning model.

2. The method for studying the fire resistance limit of steel-tube confined concrete after impact according to claim 1, characterized in that: The specific steps of S1 are as follows: a. Establish a static analysis model using the finite element software ABAQUS, obtain the initial bearing capacity of the component by applying displacement, obtain the axial load value according to the axial compression ratio, and apply the axial load in the impact model and fire resistance limit model. b. Establish an impact model using explicit dynamics. By assigning mass to the impacting body, setting the distance between the impacting body and the component, and setting the velocity of the impacting body in a predefined field, the impact event can be simulated. c. Set up the Coulomb friction model of the steel pipe-concrete interface and the impactor-steel pipe interface; d. Divide the continuous geometric model into many small meshes; e. Determine the failure displacement of the component based on the maximum displacement criterion, and obtain the critical value of the maximum impact energy that the component can withstand.

3. The method for studying the fire resistance limit of steel-tube confined concrete after impact according to claim 2, characterized in that: Before the S2 is based on the ISO-834 standard heating curve, the heating situation of the component under the ISO-834 standard heating curve is calculated. The heating process and node temperature of the component are obtained by using the heat transfer analysis step, surface heat exchange conditions, surface radiation and heat conduction.

4. The method for studying the fire resistance limit of steel-tube confined concrete after impact according to claim 2, characterized in that: When establishing the initial fire resistance limit model in S3, a model with consistent component size and mesh division with the temperature field model is established. The static general analysis step is adopted, the load is applied according to the axial compression ratio, and the temperature field calculation results are imported into the temperature predefined field. The "displacement-time" curve extracted from the load point is used as the fire resistance limit result.

5. The method for studying the fire resistance limit of steel-tube confined concrete after impact according to claim 1, characterized in that: The steps for calculating the fire resistance limit of the components in S5 are as follows: The damaged state after impact is used as the initial state and imported into the fire resistance limit model; Import the impact-damaged state as the initial state into the predefined field of the Initial analysis step, and import the impact-damaged state as the initial state into the Initial predefined field of the initial fire resistance limit model. Import the calculated temperature field model results into the predefined temperature field of the Step-1 analysis step; Step-1 analysis imports the temperature field results into the predefined field and applies a load in Step-1 analysis to obtain the fire resistance limit value after impact.

6. The method for studying the fire resistance limit of steel-tube confined concrete after impact according to claim 1, characterized in that: The mesh generation of the post-impact fire resistance limit model in S5 is consistent with that of the impact model and the temperature field model, and the boundary conditions are consistent with those of the impact model.

7. The method for studying the fire resistance limit of steel-tube confined concrete after impact according to claim 1, characterized in that: The method for establishing the machine learning model in S6 is as follows: Using the stacked regression method, three CatBoost models were used as base learners, and the PSO particle swarm optimization algorithm was used for hyperparameter tuning. The LightGBM model was used as the meta-learner. The importance of parameters and the coupling effect of key parameters were analyzed by combining SHAP to build the machine learning model.

8. The method for studying the fire resistance limit of steel-tube confined concrete after impact according to claim 7, characterized in that: In the S6 step, the SHAP analysis calculates the SHAP value of each feature in all samples, takes the mean of the absolute SHAP values ​​of each feature, and sorts them from high to low according to the mean to obtain the feature importance ranking.

Images